Fire alarming method based on video detection

Abstract

The invention discloses a fire alarming method based on video detection. Aiming at video images obtained by a monitoring camera in a fixed view field, the fire alarming method includes the steps of steps 1, carrying out binaryzation division on images in an Ostu method to obtain a foreground area, step 2, removing noisy points through morphological filtering, step 3, summarizing various characters like color brightness, geometrical shapes, diffusion property, irregularity and flicker frequency of the foreground area, and step 4, carrying out synthetic judgment on the computed various characters of the foreground area so as to judge whether flame exists in the video images. A large number of interference influences on binaryzation division of images can be reduced in the Ostu method. Meanwhile, aiming at the defect that a commonly-used flame detection method in the prior art is single or insufficient in judgment character, the fire alarming method extracts the various characters of the foreground area obtained in the Ostu method for synthetic judgment, therefore, quick and accurate recognition on whether the flames exists in the video images is achieved, and meanwhile error rates are reduced greatly.

Description

technical field [0001] The invention belongs to the field of pattern recognition, specifically relates to the technical field of fire monitoring, in particular to a video image pattern recognition method based on feature analysis to detect fire phenomena. Background technique [0002] The prevention and detection of fire has always been the goal pursued by human beings in the process of fighting against fire. For the indoor environment, smoke-sensing, temperature-sensing, and light-sensing detectors can be installed, which use the characteristics of smoke, temperature, and light of fire flames to detect fires. But for a long time, in tall spaces or outdoors, early fire detection has been a difficult problem worldwide. Because in this type of environment, there are many factors that affect fire detection, mainly including: detection method, space height, heat barrier, coverage, air velocity, explosive / toxic gas, acceptable false alarm rate, alarm information management And ...

AI Technical Summary

Problems solved by technology

Some existing video flame detection methods, either because of weak anti-interference ability, or because the discriminant features used are relatively simple, make it difficult to avoid false alarms and false alarms, and their adaptability and stability in complex monitoring environments are also not good. satisfactory

Images